Diagnostics for Periodically Operated Actuators

Abstract

Increasing constraints on quality, reliability and minimum downtime require the revision of existing maintenance approaches. Preventive maintenance, or even more reactive maintenance, require information about a system’s condition in order to enable predictive maintenance approach. Condition monitoring requires efficient sensing and data processing for extraction of condition-related signal features. Advances in both connectivity and embedded systems enable a wide range of possibilities in the field of condition monitoring.

This thesis develops signal processing tools and hardware solutions optimized for, but not limited to, diagnostics of periodically operated actuators. These actuators are mechanical or electromechanical systems that experience non-uniform loads during an operating cycle. The platform presented in this thesis serves state-of-the-art vibration and acoustic measurements and combines the quality of high-end acquisition systems with the portability of IoT devices. This allows for temporary field installations and monitoring of critical industrial equipment. Cyclostationary analysis enables diagnostics based on signals with strong random components by extracting modulation signatures otherwise unattainable by conventional time or frequency domain analysis, as demonstrated with applications to diaphragm pumps and cutting tools. An extension to the Integrated-Electronics-Piezoelectric (IEPE) industry standard for vibration measurements stretches the applications to a wide range of measurands like temperature, pressure or mechanical strain. These stretched capabilities enable a more unified sensing strategy and decrease complexity of the condition monitoring systems; thus, it further supports miniaturization and on-the-edge applications.